MindStar: Enhancing Math Reasoning in Pre-trained LLMs at Inference Time
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914849044299776 |
|---|---|
| author | Kang, Jikun Li, Xin Zhe Chen, Xi Kazemi, Amirreza Sun, Qianyi Chen, Boxing Li, Dong He, Xu He, Quan Wen, Feng Hao, Jianye Yao, Jun |
| author_facet | Kang, Jikun Li, Xin Zhe Chen, Xi Kazemi, Amirreza Sun, Qianyi Chen, Boxing Li, Dong He, Xu He, Quan Wen, Feng Hao, Jianye Yao, Jun |
| contents | Although Large Language Models (LLMs) achieve remarkable performance across various tasks, they often struggle with complex reasoning tasks, such as answering mathematical questions. Recent efforts to address this issue have primarily focused on leveraging mathematical datasets through supervised fine-tuning or self-improvement techniques. However, these methods often depend on high-quality datasets that are difficult to prepare, or they require substantial computational resources for fine-tuning. Inspired by findings that LLMs know how to produce the right answer but struggle to select the correct reasoning path, we propose a purely inference-based searching method -- MindStar (M*). This method formulates reasoning tasks as searching problems and proposes two search ideas to identify the optimal reasoning paths. We evaluate the M* framework on both the GSM8K and MATH datasets, comparing its performance with existing open and closed-source LLMs. Our results demonstrate that M* significantly enhances the reasoning abilities of open-source models, such as Llama-2-13B and Mistral-7B, and achieves comparable performance to GPT-3.5 and Grok-1, but with substantially reduced model size and computational costs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_16265 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | MindStar: Enhancing Math Reasoning in Pre-trained LLMs at Inference Time Kang, Jikun Li, Xin Zhe Chen, Xi Kazemi, Amirreza Sun, Qianyi Chen, Boxing Li, Dong He, Xu He, Quan Wen, Feng Hao, Jianye Yao, Jun Machine Learning Although Large Language Models (LLMs) achieve remarkable performance across various tasks, they often struggle with complex reasoning tasks, such as answering mathematical questions. Recent efforts to address this issue have primarily focused on leveraging mathematical datasets through supervised fine-tuning or self-improvement techniques. However, these methods often depend on high-quality datasets that are difficult to prepare, or they require substantial computational resources for fine-tuning. Inspired by findings that LLMs know how to produce the right answer but struggle to select the correct reasoning path, we propose a purely inference-based searching method -- MindStar (M*). This method formulates reasoning tasks as searching problems and proposes two search ideas to identify the optimal reasoning paths. We evaluate the M* framework on both the GSM8K and MATH datasets, comparing its performance with existing open and closed-source LLMs. Our results demonstrate that M* significantly enhances the reasoning abilities of open-source models, such as Llama-2-13B and Mistral-7B, and achieves comparable performance to GPT-3.5 and Grok-1, but with substantially reduced model size and computational costs. |
| title | MindStar: Enhancing Math Reasoning in Pre-trained LLMs at Inference Time |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2405.16265 |